JP2002521082A - Synthetic structural imaging and volume estimation of biological tissue organs - Google Patents

Abstract

A system (10) having a processor and memory generates a low frequency ultrasound signal to be applied to tissue (14) to generate a weighted sum of ramps, steps and impulse signatures of the tissue. Occurs. The system (10) analyzes the tissue signature to determine a low frequency target profile. This low frequency target profile is used to generate a graphical representation of the tissue using the set of stored tissue data as well as to estimate tissue capacity and to classify the tissue with respect to tissue type and condition. The classifier may include a neural network (34) and / or a nearest neighbor processor (36). The system (10) implements the imaging system and method as a non-invasive acoustic measurement. The system and method uses synthetic structural imaging (SSI) technology to provide unique information about the size and shape of biological structures for the classification and visualization of normal and abnormal tissues, organs, biological structures, etc. I do.

Description

DETAILED DESCRIPTION OF THE INVENTION

[Background] [1. TECHNICAL FIELD The present disclosure relates to medical imaging, and more particularly to systems and methods for imaging and determining body organs.

[2. Description of Related Art] Internal anatomy of organ system and supporting vasculature
Non-invasive visualization of the neural network provides valuable medical diagnostic information for the patient. Direct volume rendering and surface-fitti
ng) algorithm using the volume visualization (volume) representing the anatomy
There has been considerable research over the past decade exploring the technology of visualization. These volume visualization techniques have been applied to various imaging methods such as ultrasound (US), magnetic resonance imaging (magnetic resonance imaging) (MRI) and computed tomography (CT).

[0003] In the field of diagnostic ultrasound, the use of conventional 2D images requires the operator to attempt to mentally reconstruct and visualize the 3D characteristics of anatomy and related pathology. . However, the ability to "think 3D" varies considerably among clinicians and depends on their experience and innate ability in spatial perception. It is sometimes very difficult for radiologists to generate 3D “images” from 2D slices, detect some lesions even with multiple views, and visualize the supporting vascular system.

[0004] Three-dimensional ultrasound (3D US) images are derived from two-dimensional successive slices from a regular ultrasound scan. Tissue volumes are spatially sampled, stored digitally, and simultaneously displayed in a multiplanar array form, rotated, thresholded, and dissected as needed to optimally view the structure of interest. Provide any desired three vertical anatomical planes using an electronic scalpel.
By retaining the full volume of data, the analysis can be performed offline after the patient leaves the clinic. This allows multi-plane images to be explored in many arbitrary planes and using various processing options. For example, the analysis may consider obtaining specific statistics of the region of interest and their change over time, combining information from multiple methods, and describing and compensating for motion.

[0005] 3D imaging has been very effective in obstetric and gynecological (Ob / Gyn) studies. It is used to detect congenital malformations in fetal features such as cleft lip and palate (10 to 39 weeks gestation) and for early detection of chromosomal abnormalities in the first trimester (trimester). Cystic swelling and physiological head transparency (
physological nucai translucency). 3D US has also enabled the measurement of fetal organ volume for assessment of fetal growth and fetal abnormalities. First, 3D
The US has measured the fetal lung volume and has enabled the possibility of relating it to gestational age and fetal weight. The 3D multi-plane US may also be effective in identifying and evaluating standard heart planes from 14 weeks to maturity. Clinical trials have shown that 3D measurements of cardiac volume can be used to improve screening for fetal cardiac malformations and have the best results during gestational weeks 22 and 27.

[0006] 3D US also describes ventricles, uterine anatomy, carotid arteries and internal lumens (endo).
luminal structures were effective in improving the visualization of blood vessels and tumors in the prostate and breast. In recent clinical prostate studies, 3D
The US has issued a permanent transperineal seed insert for the goal of real-time optimization of prostate brachytherapy.
Gives a greater level of confidence in identifying a dioactive seed implant. In breast studies, intraoperative 3D US is very effective in detecting and locating free silicone from a ruptured breast insert when the ruptured breast insert is surgically removed. 3D US locates the more accurate spatial location of the lesion for biopsy compared to regular 2D US and provides an accurate assessment of the vasculature and their associated pathology. In extensive clinical studies, 3D internal lumen US could not be obtained using normal 2D imaging, anatomy such as vessel lumen size and shape, plaque distribution, location and type. It provided unique information about the spatial relationship of the dynamic structure. 3D US
Can give a more accurate distribution of the tumor along the ureter, an adjacent relationship to its structure, and give a measure of tumor volume. It is also a colorectal mass (col)
The visualization of the original mass and the determination of the progress can be greatly facilitated.

[0007] Some major limitations of 3D ultrasound imaging are: (1) the substantial number of "looks" or "slices" required for image reconstruction (usually magnetic resonance imaging (M
(500-600 slices for RI) and computed tomography (CT) and about 64 slices for ultrasound), (2) long data acquisition time required for image formation, (3)
The accuracy required for multi-plane anatomical registration (eg, a resolution of about 0.5 mm or less), and (4) requirements for large memory storage and rapid calculations. There is a need for improved imaging techniques to address such limitations. For example, reducing the acquisition time greatly minimizes the effect on target movements, such as fetal movements, and results in less exposure time, and concomitantly from normal biological activity or sudden movements. Biological effects from
Brings you to less risk.

[0008] Other imaging techniques have been used to improve object detection and classification. For example, Synthetic Structural Imaging
(SSI) technology uses low frequency transmission for successful detection and classification of both aircraft and acoustic mines and submarines, both radar and sonar.

The concept of SSI has been shown to provide acoustic target identification and structural feature estimation in sonar applications. Test results have shown that the acoustic transient response exhibits unique characteristics of target identity with features strongly related to simple geometric features. It has shown that such a signature can be used to provide a sufficiently high quality image information as well as a sufficiently accurate capacity estimator for a narrow band for nearly accurate target identification.

[0010] The SSI employs a ramp response analysis, which was developed in an air wing identification radar study and has been successfully applied to imaging underwater (scaled) targets. Similar to normal high frequency imaging, SSI methods are direction dependent, but quite robust. It has lower resolution than the prior art, but the correlation to its shape is much stronger. SSI technology has been shown to be very promising for certain radar and sonar applications in previous experimental studies.

[0011] The application of SSI technology to biological media may provide an estimate of the volume of biological organs, tumors and other structures. To date, primary tissue classifiers for volume, size and shape in a clinical environment.
There has been little progress in estimating fiers) and relating them to histopathology. Furthermore, discrimination of normal from abnormal tissue has not been successfully achieved using SSI.

Overview A novel non-invasive technique that uses SSI technology to provide unique information about the size and shape of biological tissue structures for the classification and visualization of normal and abnormal tissues, organs, tumors, etc. Acoustic measurement and imaging systems and methods are disclosed.

[0013] The SSI system includes a processor and memory for generating a low-frequency ultrasound signal to be applied to a biological structure to generate a composite structural image of the structure. S
The SI system analyzes the low frequency ramp response of the tissue structure, uses the low frequency ramp response to generate a graphical representation of the tissue structure as well as estimate the volume of the tissue structure, and provides a set of stored tissue data. Is used to classify the tissue structure with respect to tissue type and status. The classifier may be a neural network and / or a nearest neighbor (near neighbor) rule.
est neighbor rule) processor.

The disclosed systems and methods utilize low frequency ultrasound transmission for detection and classification, where amplitude and phase information as a function of tissue type, target direction and frequency are stored as an acoustic database. You. The system utilizes the correlation between the target shape and low frequency signature features.

[0015] Low frequency imaging combined with high frequency imaging requires significantly fewer imaging planes or "slices" than conventional methods to achieve real-time 3D imaging of tissue structures. The systems and methods provide unique measures of biological tissue volume as well as material composition, which can be used as inputs to a classifier for tissue classification. Predetermined tissue-specific signal waveforms, a priori information about the general properties and anatomical location of the biological "target" of interest, and temporal and frequency processing minimize possible ambiguities and artifacts Used for The disclosed system and method provides for color flow imaging.
g) Can be integrated with existing high-end radiology or cardiology imaging systems, including technology.

[0016] Features of the disclosed tissue imaging system and method will become more readily apparent and better understood by referring to the following detailed description of an exemplary embodiment of the invention, taken in conjunction with the accompanying drawings. Will be.

DESCRIPTION OF THE PREFERRED EMBODIMENTS Referring now particularly to the drawings, in which like reference numbers identify similar or identical components, and as shown in FIG. Normal and abnormal tissues,
A tissue imaging system and method for determining the size and shape of a biological structure for classification and visualization of organs, biological structures, etc., is described.

For clarity of explanation, the exemplary embodiments of the disclosed tissue imaging system and method are presented as having individual functional blocks, which are referred to as “processors”.
Or it may include a functional block labeled "processing unit". The functions represented by these blocks may be provided through the use of shared or dedicated hardware, including but not limited to hardware capable of executing software. For example, the functions of the processors and processing units presented herein include:
It may be provided by a shared processor or by a plurality of individual processors. Furthermore, the use of functional blocks with the accompanying labels herein should not be interpreted as referring exclusively to hardware capable of executing software. The exemplary embodiment stores digital signal processor (DSP) hardware, such as an AT & T DSP16 or DSP32C, a read-only memory (ROM) for storing software for performing the operations described below, and stores DSP results. Random Access Memory (RAM) for Very large scale integration (VL
SI) Hardware embodiments may also be provided, as well as custom VLSI circuits in combination with general purpose DSP circuits. Any and all of these embodiments may be considered to fall within the meaning of a label for a functional block as used herein.

In the exemplary embodiment of FIG. 1, system 10 processes an input data signal provided by a plurality of sensors 12. The plurality of sensors 12 respond to low frequency ultrasound signal insonification during testing, for example, a transmitter 46 and an ultrasound generator such as the projector 16 (both in the art). About 10 kHz, having an ultrasonic waveform generated and transmitted by
Responsive to biological tissue 14 in response to frequencies ranging from z to about 1.0 MHz. The projector 16 transmits a wide range of carefully controlled ultrasonic signals during the test to the transmitter 46.
Can be applied to the biological tissue 14 to generate a corresponding lamp response signature. The ramp sign is detected by a sensor 12, which generates a corresponding received input data signal. The received input data signal is then analyzed to determine characteristics of the biological tissue 14, such as size, shape, composition, volume, and normal or abnormal conditions.

The received input data signal is processed by a preamplifier 18 and then a filter 20
Is filtered by The filtered data signal is then processed by a processing unit. The processing unit includes a central processing unit (CPU) 22 that operates with a data acquisition and control logic card 24 and a DSP card 26. CPU 22 and other components of system 10 may be controlled by, for example, an application program written in the C ++ programming language to implement the features and methods described herein.

The memory may be hard drive 28 and / or RAM, and
RAM on the P-card 26 may be used for faster memory access and processing. The hard drive 28 and the cards 24, 26 communicate with the CPU 22 using a bus 32, such as a PCI bus, running a PCI protocol. CPU
22 may include a neural network 34 for classification and / or a nearest neighbor (NNR) processor 36, as described in more detail below.

After processing the input data signal, the system 10 outputs the data to an output device such as a printer 38 or, optionally, a display 40, an audio system, or other type of output device known in the art. To generate a processed signal. The output device may output an alphanumeric text message indicating the condition of the tissue 14 being tested, and / or to some extent that the tissue under test 14 is within or outside of the predetermined normal tissue condition. May output a categorization message indicating, for example, 100%
The percentage compared to normal can be generated or output. The output device may also generate a video or graphic display of the tissue 14 based on the processing of the tissue lamp sign. For example, display 40 includes screen 42 for displaying graphical display 44 to a clinician.

The CPU 22 sends a control signal to the transmitter 46. Note that the transmitter 46 includes a programmable waveform generator 50 that generates a signal waveform, and includes a power amplifier 48 that amplifies such a signal waveform. The signal waveform is sent to a projector 16 that generates an ultrasonic wave applied to the tissue structure 14.

In the exemplary embodiment, the projector 16 is a piezoceramic
amic) projector, the piezoceramic projector comprising one or more transducer piezoceramic components and insonifying the tissue structure 14, such as the internal organs of the patient, in a frequency range from about 10 kHz to about 100 kHz.
calibrated to insonify). Projector 16 is a Naval Research Laboratory (Na
val Research Laboratory (NRL) Underwater Sound Reference Division
) (USRD) may be an F30 or F41 transducer. The return of the echo from the tissue structure is received by the sensor 12. Note that the sensor 12 may be four calibrated broadband sensors, such as the B & K model 8103, oriented to provide four distinct target aspects in an orthogonal plane.

In an alternative embodiment, sensor 12 and projector 16 insonify tissue structure 14, such as a breast tumor, and provide a backscattered return of about 100 kHz.
And receiving in the frequency range of about 800 kHz from a single device, including three custom piezoceramic transducers designed and manufactured. Highly crystalline oriented types of thermoplastic polymers such as polyethylene terephthalate are also available from about 10 kHz to about 1 MHz.
Up to a wideband frequency response. The outputs of the sensor 12 and the transducer are sent to individual preamplifiers 18 and anti-aliasing filters 20 via coaxial cables, and then to C
It is sent to a data acquisition card 24 operatively connected to the PU 22. In addition,
CPU 22 may be embodied as a personal computer or workstation.

The preamplifier 18 provides an input dynamic range of about 80 dB while minimizing noise and distortion, such as the AD601 commonly used in medical ultrasound, preceded by a low noise JFET. A separate and independent low noise wideband programmable gain amplifier. The preamplifier 18 is configured on a computer card or board that plugs into the system 10 and allows the user to change the gain setting via software control without degrading the frequency response when increasing the gain. Can be done. Filter 20 provides about 84 to provide aliasing protection.
Anti-aliasing, such as a TTE delay-equalized elliptic filter, followed by a preamplifier 18 to provide a relatively high roll-off rate of dB / octave.
It can be a filter. In an exemplary embodiment, the receiver preamplifier 18 has an input dynamic range of about 72 dB while minimizing noise and distortion.

The output of the anti-aliasing filter 20 is sent to a data acquisition and control logic card 24 or board. It should be noted that the data acquisition and control logic card 24 or board operates as a data acquisition subsystem having, for example, a throughput data rate of about 80 megabytes / second, Burr-Brown ADS802.
Or four different input S / H amplifiers, such as AD9042 from Analog Devices, and a 12-bit 10 MHz analog-to-digital converter (ADC).
). Each channel of the ADC has a full voltage range of the ADC and about 100 dB.
May have its own programmable gain amplifier with sufficient gain to provide a common mode rejection ratio of.

The data acquisition subsystem, including CPU 22, cards 22-28, and operating and control software, is based on the Sonoran Microsystem.
ms Inc. Embedded in the “FALCON” computer system available from Hi-Techniques Inc. HT-6 available from
00 "computer system.

The CPU 22 can be an “Intel” based “Pentium” microprocessor and the DSP card 26 can be a quad TMS220C6201. Hard drive 28 may have one or more Se for all storage.
gate 18.2 GB high speed SCSI hard drive.

The data acquisition and control logic card 24 formats the data into a standard personal computer file format, such as an ASCII data format, and thus uses modified system software and / or SPLUS Alternatively, the data can be reproduced in a laboratory using third-party commercial analysis software, such as an application program including MPLE. Real-time performance requires multiple COTS for DSP card 26
Achieved by using a DSP board. Using the DSP card 26, the data is acquired, the data is packed for storage in the hard drive 28 and passed to the CPU 22, while at the same time bandpassing the data low-pass filter and bandpassing the data. Decimates and performs various processing operations such as data normalization, fast Fourier transform (FFT) analysis and parameter estimation.

The system 10 determines a three-dimensional image of a biological organ that requires a minimum number of “looks”, for example, at most three slices. The system 10 also generates diagnostically useful estimates of organ volume and tumor size, and the neural network 34 and / or N
The classification of the tissue 14 using the NR processor 36 provides an estimate of the biological tissue composition. In performing the classification, a predetermined 3D STIC image of a normal tissue structure, such as a tumor-free biological organ, as well as an estimate of organ volume, for example, to train the neural network 34, and And / or used by the NNR processor 36 to compare the current tissue data with the tissue data stored in the library 21 and as a basis for classification.

The DSP card 26 is stored on the hard drive 28 using 3D STIC ultrasound data acquisition, signal processing, and image reconstruction techniques derived from in vivo and in vitro analysis of predetermined and identified tissue. Generated organization database.

Synthetic structural imaging (SSI) utilizing low frequency ramp response signatures offers a unique and effective technique for presenting three-dimensional medical data to clinicians. By applying SSI technology with advanced signal and image processing methods, system 10 obtains clinically meaningful measurements of the size and shape of biological organs, and their composition and condition.

The system 10 applies a low frequency signal insonification (ie, ramp sign) that is matched to the spatial frequency of an anatomical structure, such as tissue 14,
There the low frequency gives unique information about the overall size and proper shape of the tissue 14. The system 10 then reconstructs a 3D image of the tissue phantom without using more than three individual "looks" or insonifying planes and with data acquisition times approaching real-time operation.

Estimation of the volume of target tissue, such as organs and tumors, is performed by determining the volume of target tissue as a unique and spatially invariant classification parameter derived from the processed low frequency echo returns. Is done.

In use, the system 10 measures critical dimensions of various organs, particularly the breast, prostate, uterus, and testes, for the purpose of detecting pathology prior to performing a biopsy. The system 10 can also be applied to provide unique anatomical information of eyes, fetal head growth and ventricles, tumors and other lesions.

In another embodiment, the system 10 combines SSI technology using low frequency imaging with known high frequency imaging (eg, using normal ultrasound frequencies in the 2 to 12 MHz region), thereby Only a very small number of imaging planes or "slices" are required to produce a meaningful 3D diagnostic image. In an exemplary embodiment, up to three slices are used as compared to at least 64 slices in the prior art. The unique information provided by the SSI, along with a reduction in data acquisition time, is useful in extensive tissue studies, especially in ultrasound aortic examination, in intraoperative procedures, in the analysis of specific body organs such as the prostate and kidneys. ,
And facilitates and significantly improves clinical interpretation in ophthalmology.

The ramp response signature is the basis for low-frequency characterization, where the derived physical-optical approximation of the target lamp response R (t) is the target cross-section A (r) along the direction of propagation of the incident field. And has a characteristic that can be expressed as:

[0039]

(Equation 1)

Here, c is the propagation velocity in the medium, and r is the radial distance. Thus, the lamp response provides a unique low frequency measure of target shape, orientation and material. The classical acoustic target backscatter response versus kA is shown in FIG. 2 for the Rayleigh, resonance and optical regions, with the Ka region indicated for SSI operation. Here, k is a wave number, that is, a frequency, which is 2π / λ, λ is a wavelength, and A is a target radius. From previous experiments with radar and sonar tests, a valid estimate of the lamp response was found to be the insonification frequency located in the upper Rayleigh region, and the resonance region with low target scattering properties, ie, about 0.8 kA shown in FIG. From about 30kA
Are obtained for the region 52 up to. Various information may be derived from the time and frequency domain analysis of region 52. Using time domain analysis, synthetic image generation and target parameters such as target area, volume, length, diameter and aspect, ie, 3
An orientation determination in D space may be performed. Using frequency domain analysis,
Feature vectors such as Doppler characteristics, target aspect, spectral shape and misclassification probabilities can be generated. The spontaneous resonance of the target can also be determined from frequency domain analysis, which facilitates determining a type of target such as liver tissue as opposed to bone tissue.

Low frequency imaging is characterized by a narrow band and low absorption loss, while high frequency imaging is characterized by a wide band and high absorption loss. Thus, high frequency imaging tends to be applied to shorter tissue depths for characterization. High frequencies characterize the fine details of the target, while low frequencies give information about overall dimensions and proper shape. High frequencies can be used to sharpen the image, but images are difficult to achieve without low frequency information.

In electromagnetic applications, the physical optics approximation provides an estimate of the waveform-target size and shape for the illuminated portion of the target, which is significant and smooth if the target is a perfect conductor. And has a size of about several wavelengths.
Test results show that the lamp response can be approximated by examining the target's electromagnetic response over a frequency range corresponding to wavelengths starting at half the target and increasing to about ten times the target's dimensions.

It has been found that the lamp response is also applicable to ultrasound imaging. System 1
The imaging technique used by O.0 employs low frequency ultrasound signals for the target size and shape, and such imaging utilizes high frequency short pulse data for structural discontinuities. Augmented by providing additional information.

As shown in FIG. 2, ultrasound having a low frequency ramp response may be applied to the tissue 14. Such a ramp response may be represented by a received echo signal that may vary over time and may be discontinuous. As described herein, low frequencies may be used for tissue 14 detection and classification. This ambiguity employs short high frequency pulses using an impulse response, despite the fact that there may be a many-to-one correspondence between the ramp response features and the possible structural discontinuities that may produce them. Can solve the problem. Thus, the lamp response from low frequency pulses can be discerned by using high frequency short pulses. The target impulse response is sensitive to curvature in the cross-sectional area, and is therefore sensitive to boundary discontinuities and scattering centers. Therefore, by including high frequencies to define the target scattering center,
The number of low frequencies required to image the target can be significantly reduced. This means
Suggests that the optimal target response is a weighted sum of the ramp, step and impulse responses.

In another embodiment, the low frequencies that generate the ramp response are used as feature vectors for pattern classification by neural network 34 and / or nearest neighbor (NNR) processor 36. As an input to the neural network 34, the ramp response is
A training set of tissue data (tra
Generate a neural network output that classifies the ramp response with respect to the ining set. The nearest neighbor rule, as implemented by the NNR processor 36, is generally a strong decision rule that is ideally suited for discriminating low frequency data. Radar test results can be achieved with about four frequencies of the ramp response, with greater than 90% reliability of the classification utilizing amplitude information and vertically polarized data. Fewer frequencies are required by employing phase information. This is because phase is a sensitive measure of change in target shape. In acoustic applications, particle velocities are "rotary (r
ot ", only amplitude and phase modulated data is needed.

Heretofore, low frequency insonification has not been widely used for biological analysis and diagnosis. One low frequency diagnostic technique, using frequencies in the range of about 10 to about 1000 Hz, involves mechanically vibrating the tissue at a low frequency that modulates the ultrasound carrier frequency, resulting in a so-called sonoelasticity in the tissue. Abnormal local elasticity can be imaged. The resulting Doppler displacement is a function of tissue stiffness, i.e., Young's modulus, and is therefore known in the art as color flow mapping.
and can be displayed with a Doppler flow mapping imaging system.

A preliminary rule of thumb is that “hard” or “hard” tissue vibrates less than soft tissue, depending on the degree of firmness, and that vibration frequencies between about 100 and about 300 Hz Because it is useful.

The low frequencies required to apply SSI technology are generally higher than the frequencies used for acousto-elastic or elastography. SSI
Gives an estimate of the classification parameters that were not previously derivable,
It then uses system 1 to derive feature patterns representing benign and malignant diseases.
Used by 0.

Acoustic echo signatures are generally rich in information from distributed tissue structures, and by using SSI techniques for acoustic and elastic scattering in real biological media, Structures can be detected and classified with substantial accuracy.

The system processing can be used to determine the frequency dependence of tissue attenuation, the response of tissue shear wave generation, and the effects of interconnected tissue and adjacent structures, the effects of veins and arteries, and broad beam insonification. Take into account.

Ultrasonic attenuation increases with increasing frequency and increasing tissue penetration depth. Because of the significant variation (variance) in the response due to the damping with respect to the overall dynamic range of the lamp response, the variation (variance) adversely affects the relationship between the lamp response and the organ geometry. The frequency range required to apply SSI to fit the organ of interest is about 10 to about 100 kHz. About 1.0dB / cm-MH
For one-way longitudinal absorption of z, the two-way absorption received at a depth of about 10 cm is:
It is about 0.2 to about 2.0 dB over the frequency band.

Such fluctuations (dispersion) can be compensated for in the transmitted ultrasound signal by having a dynamic range of about 48 dB. Based on the actual human tissue organs considered, the adopted SSI frequency may also generate an oscillatory shear mode. In fact, some cross-coupling of modes between shear and compression can occur, where system 10 evaluates such shear waves. Typically, the shear wave attenuation coefficient is
It was measured to be about 10 ⁴ times the longitudinal wave attenuation coefficient.

In theory, the biological ultrasound ramp response is affected by tissue / organ deposits such as blood vessels, cartilage, and adjacent anatomical structures. However, at the spatial frequencies required for synthetic structural imaging of very large organs, the vessels involved are generally acoustically transparent.

Since normal SSI operation employs a wide beam width, some organs can be insonified at a time. To decompose the echo ramp signature into components associated with individual organs, the system 10 uses a priori information regarding the general spatial properties of the organ, selection of target aspects, and temporal gating. High frequency 2D image data can be used to resolve any ambiguity in identification and classification. Thus, the system 10 can obtain the size and shape of biological organs and tumors and determine their composition.

Prior to use, system 10 derives empirical 3D images and measures of volume and material composition using in vivo and in vitro measurements of low frequency ramp responses of organs and tissues, including tissue mimic breast tumors. And configured to store them in the library database 21 in the hard drive 28. Such data from tissue organs can of course be corrupted by tissue plaques, system noise and artifacts,
They can be introduced by tissue / organ attachments and adjacent structures. System 10
Is integrated into known imaging systems and can be used to perform in vivo tests on human subjects to enhance and refine the library database 21.

The system 10 can be used to empirically obtain ramp responses for specific biological organs and tumors, generate low-frequency composite images of biological organs and tumors, and estimate organ and tumor volumes. To derive measures of tissue composition from the measured ramp response, such as density, elasticity, to evaluate other SSI biological structural properties such as damping and target aspect, and The signal transmitted from the projector 16 is used to evaluate the effect of the beam insonification.

From empirical data collection, unique signal waveforms or ultrasound signatures corresponding to specific tissue organs and tumors of interest are stored on hard drive 28. The signal waveform is designed such that when transmitted, the signal waveform generates a tissue phantom having distinct ramp response signatures for specific tissue organs such as prostate, kidney, eye and breast tumors. Signal waveforms that include both amplitude and phase modulation over the required frequency band can be adjusted to match the spatial frequencies that characterize these structures.

For example, the system 10 may provide a signal waveform having a frequency of about 10 to about 100 kHz for organ visualization, and about 100 kHz to about 5 kHz for visualizing a breast tumor at a size of about 2 mm to about 5 mm. Signal waveforms having frequencies up to 800 kHz may be used. For such organ detection and identification, the fundamental to harmonic ratio of the transmitted signal is about 48 dB, and the signal used is a predetermined complex function to provide sufficient dynamic range and minimal error echo. Can be sent according to

Prior to collecting the data, all system components include, for example, ultrasound spatial and frequency response including projector and hydrophonic spatial frequency characteristics and response, source level and spectral purity, reception Bandwidth, preamplifier gain and input noise level, any integral and differential non-linearities, any harmonic and IM distortion, any ADC spurious-free dynamic range, any backscatter data, and measurements Calibrated to establish any naturally occurring background transients within the specified frequency band. The amplitude and phase differences between the channels are then measured and compensated. In system 10, the power level employed for transmission is determined by the SUM specified in AIUM / NEMA standard 9-17-1981.
Compatible with PTA requirements.

The DSP card 26 performs echo data processing in which the FFT of the target complex spectral response is appropriately weighted to form an FFT approximation of the target ramp response signature. The ramp response derived from the echo amplitude and phase data is further processed to approximate a target cross section function or "profile function."
The empirically derived profile function is modified based on known geometric constraints and used as input data for the image reconstruction techniques employed by CPU 22.

Image reconstruction generates, for example, an isometric image of a target using a confined surface, where the target is generally a few, such as a shape indicated by a circular or elliptical cross-section. Including the simple shape of. Image reconstruction of such shapes typically requires only a few parameters for contour estimation. A generalized surface, such as an ellipse, is fitted to the set of profile functions, and at least one such generalized surface in the set of profile functions is assigned a look angle (look angle).
). The image is then generalized by calculating the image surface surrounding a common volume for substantially all single aspect angle limited surfaces. The orthogonality between the look angles can be used to greatly simplify such image processing.

The three-dimensional reconstructed image is a global examination (gr) of the actual tissue images in the library.
oss emission). The actual volume of each tissue phantom employed is compared to the volume measured by integrating the empirically derived profile function in various aspects. The capacitance error is used as a measure of image accuracy. Further, the lamp response is examined to derive information about the composition of the tissue.

As shown in FIG. 3, system 10 operates using a method that includes the following steps. The following steps include, in step 54, starting imaging of the tissue 14, for example, by initializing and calibrating the system 10, and in step 56, storing the tissue signature in a database in the hard drive 28. And generating a low frequency ultrasonic wave to be applied to the tissue 14 using the tissue signature in step 58, and detecting the tissue structure using the sensor 12 in step 60, A step in which the signature is generated by the tissue 14 in response to the low frequency ultrasound;
Processing the tissue ramp sign to determine tissue characteristics.

Step 62 of processing the tissue ramp sign includes, for example, steps 64-68
May be included. In an exemplary embodiment, the system 10 uses the classifier such as the neural network 34 and / or the NNR processor 36 to use the predetermined tissue data to classify the tissue 14 in step 64. Is categorized. A description of neural networks and NNR processors is provided in U.S. Pat. 09 / 167,868, the contents of which are incorporated herein by reference.

Further, the system 10 determines the capacity, aspect, shape, etc. of the tissue 14 at step 66 using SSI techniques, signal processing techniques such as FFT processing, etc., and the system 10 also determines at step 68 It is determined whether the state of the tissue 14 is normal or abnormal.

The system 10 may then output the tissue characteristics to the clinician. The system 10 may also generate a graphical representation of the tissue 14 at step 70 using such tissue characteristics.

Thus, using the systems 10 and methods described herein, a clinician can non-invasively and acoustically measure and generate images of a patient's tissue structure to produce normal and abnormal tissue. Unique information regarding the size and shape of biological structures for classification and visualization of organs, biological structures, etc. may be provided with improved accuracy and diagnostic analysis.

Although the disclosed tissue imaging systems and methods have been particularly shown and described with reference to preferred embodiments, various modifications in form and detail may be made without departing from the scope and spirit of the invention. It will be understood by those skilled in the art. Accordingly, such modifications, as suggested above, but not limited thereto, should be considered to be within the scope of the present invention.

[Brief description of the drawings]

FIG. 1 is a block diagram of a tissue image forming system.

FIG. 2 is a graph of a backscatter response.

FIG. 3 is a flowchart of an operation method of the tissue image forming system.

──────────────────────────────────────────────────続 き Continuation of front page (81) Designated country EP (AT, BE, CH, CY, DE, DK, ES, FI, FR, GB, GR, IE, IT, LU, MC, NL, PT, SE ), OA (BF, BJ, CF, CG, CI, CM, GA, GN, GW, ML, MR, NE, SN, TD, TG), AP (GH, GM, KE, LS, MW, SD, SL, SZ, UG, ZW), EA (AM, AZ, BY, KG, KZ, MD, RU, TJ, TM), AE, AL, AM, AT, AU, AZ, BA, BB, BG, BR , BY, CA, CH, CN, CU, CZ, DE, DK, EE, ES, FI, GB, GD, GE, GH, GM, HR, HU, ID, IL, IN, IS , JP, KE, KG, KP, KR, KZ, LC, LK, LR, LS, LT, LU, LV, MD, MG, MK, MN, MW, MX, NO, NZ, PL, PT, RO, RU, SD, SE, SG, SI, SK, SL, TJ, TM, TR, TT, UA, UG, US, UZ, VN, YU, ZA, ZW

Claims (20)

[Claims] 1. A projector for generating a low frequency ultrasound signal in a target tissue, and a tissue generated by communicating with the projector to interact with the target tissue structure with the low frequency ultrasound signal. A sensor for receiving an echo signature; a central processor in communication with the receiver to determine a low frequency ramp response from the tissue echo signature; and a low frequency ramp response using the acoustic database of stored tissue data. An acoustic measurement and imaging system comprising: memory graphics that generate a graphical representation of the target tissue structure. 2. The acoustical device of claim 1, further comprising a classifier in communication with said memory graphics to classify target tissue with respect to target tissue type or status using said database of stored tissue data. Measurement and imaging system. 3. The acoustic measurement and imaging system according to claim 2, wherein said classifier comprises a neural network. 4. The acoustic measurement and imaging system of claim 2, wherein said classifier comprises a nearest neighbor processor. 5. The method of claim 1, wherein the acoustic database of stored tissue data includes amplitude and phase information as a function of tissue type, tissue size, target direction, and insonified frequency. Imaging system. 6. A projector for generating an ultrasonic signal to a target tissue to generate a corresponding tissue echo signature, and detecting the tissue echo signature from the target tissue and generating a corresponding input data signal. A plurality of sensors, a preamplifier for processing the input data signal, a filter for filtering the input data signal, processing the filtered input data signal based on a set of stored tissue acoustic data, and An acoustic measurement and imaging system comprising: a central processor that generates an output signal; and a graphics structure that generates a graphical representation of a property of a target tissue from the output signal. 7. The acoustic measurement and imaging system of claim 6, wherein said central processor operates with a data acquisition and control logic card, a digital signal processor card and a storage device. 8. The acoustic measurement and imaging system according to claim 6, wherein said central processor comprises a neural network. 9. The processor of claim 6, wherein the central processor comprises a nearest neighbor processor.
An acoustic measurement and imaging system as described. 10. The acoustic measurement and imaging system according to claim 6, wherein said graphics structure comprises a printer. 11. The acoustic measurement and imaging system according to claim 6, wherein said graphics structure comprises a video display. 12. The projector according to claim 1, wherein the target tissue has a frequency of about 10 kHz to about 100 kHz.
7. The acoustic measurement and imaging system of claim 6, wherein the piezoceramic projector is calibrated for insonification in the z frequency range. 13. The acoustic measurement and imaging system of claim 6, comprising four calibrated sensors oriented to provide four distinct target aspects in an orthogonal plane. 14. The projector and the plurality of sensors configured as a single unit including at least two piezoceramic transducers designed to insonify target tissue and receive backscattered returns. An acoustic measurement and imaging system as described. 15. The method according to claim 15, wherein the backscatter return is from about 100 kHz to about 800 kHz.
15. The acoustic measurement and imaging system of claim 14, wherein the frequency range is: 16. The acoustic measurement and imaging system according to claim 6, wherein said preamplifier is a low noise wideband programmable gain amplifier. 17. A projector for generating low frequency ultrasound in target tissue, and generated by communicating with the projector to interact with the target tissue structure with low frequency ultrasound signals from the projector. At least one sensor for receiving the generated tissue echo signature, a central processor in communication with the receiver to determine a low frequency ramp response from the tissue echo signature, and a low-level processor using an acoustic database of stored tissue data. A memory that generates a graphical representation of the target tissue from the frequency ramp response
Providing an acoustic measurement and imaging system comprising graphics; initializing and calibrating the acoustic measurement and imaging system; retrieving a tissue echo signature from an acoustic database; Generating a sound wave to a target tissue; detecting a tissue echo signature using the at least one sensor; and processing the tissue echo signature to determine a tissue characteristic of the target tissue. A method for acoustically measuring and imaging a target tissue comprising: 18. The method of claim 17, further comprising determining one or more of volume, aspect, and shape of the target tissue. 19. The method of claim 17, further comprising generating a graphical representation of at least a portion of the target tissue structure. 20. The method of claim 17, wherein processing the tissue echo signature comprises classifying the tissue using a neural network using predetermined tissue data. Image forming method.